Side-effect actions, acting for a reason, and acting intentionally

Abstract

What is the relation between acting intentionally and acting for a reason? While this question has generated a considerable amount of debate in the philosophy of action, on one point there has been a virtual consensus: actions performed for a reason are necessarily intentional. Recently, this consensus has been challenged by Joshua Knobe and Sean Kelly, who argue against it on the basis of empirical evidence concerning the ways in which ordinary speakers of the English language describe and explain certain side-effect actions. Knobe and Kelly's argument is of interest not only because it challenges a widely accepted philosophical thesis on the basis of experimental evidence, but also because it indirectly raises an important and largely neglected question, the question of whether or in what sense an agent can perform a side-effect action for a reason. In this article, I address this question and provide a positive answer to it. Specifically, I argue that agents act for a reason whenever they perform side-effect actions as trade-offs. Thus, I claim that there are three distinct types of rational action: actions performed as ends in themselves, actions performed as means to further ends, and side-effect actions performed as trade-offs. Given this multiplicity of types of rational action, the question of whether or not actions performed for a reason are necessarily intentional is in need of refinement. The more specific question that lies at the heart of this article is whether or not side-effect actions performed as trade-offs are necessarily intentional. I conclude that, contrary to what Knobe and Kelly suggest, the question remains open.

Keywords:

• actions performed for a reason
• intentional actions
• side-effect actions
• reason explanations
• trade-offs

Acknowledgements

This article was written with the financial support of a research grant (#A00234) from the National Research Foundation of Korea. I thank two anonymous referees of this journal for helpful comments on the earlier drafts of this article.

Notes

Cushman and Mele (2008, 179) define side-effect actions as follows: X is a side-effect action performed by an agent S if and only if S successfully seeks to perform an action A, E is an effect of his doing so, X is his bringing about E, and X has the following properties: S is not at the relevant time seeking to X either as an end or as a means to an end, and X is not in fact a means to an end that X is seeking at the relevant time.

Adams and Steadman (2004) raised a similar objection in response to one of Knobe's earlier experiments (Knobe 2003) in which subjects were asked to indicate the degree to which certain sentences ‘sound right/wrong’.

For this and the other two studies reported in this article, data were collected by means of an on-line survey tool. To help ensure that subjects were competent in the English language, invitations to participate in the surveys were directed toward students attending three North American universities, where English-language proficiency is an admission requirement. Each survey targeted a different group of students, and no subject participated in more than one of these three surveys. In each case, participation was voluntary, anonymous, and uncompensated.

The follow-up paper, Knobe (2007), was written after, but published before Knobe and Kelly (2009).

While one may notice certain parallels between the claim being advanced here and the doctrine of double effect, it is even more important to note the differences between them. The doctrine of double effect is a moral claim or hypothesis about the conditions under which side-effect actions that are considered harmful (or otherwise bad) are morally permissible. Unlike the doctrine of double effect, the claim being advanced here is not a moral claim; it says nothing about the permissibility of side-effect actions. It is rather a semantic claim or hypothesis about the conditions under which side-effect actions are done for a reason.

An alternative formulation for (6c) is the following: ‘The terrorist believed that the considerations in favor of saving his son outweigh the considerations against saving the Americans’. Whether or not this alternative formulation is in any way preferable to (6c) is debatable, but it is not a debate I will pursue in the present context.

To get a sense of levels of agreement/disagreement among English speakers with the sentence ‘The terrorist intended to save the Americans’, I conducted an experiment in which subjects were asked to read the Terrorist Scenario and then evaluate that sentence on a 7-point scale from −3 (disagree) to +3 (agree). A total of 58 subjects participated in the experiment, and the average score was −2.3, which represents very strong disagreement with the sentence. This result is consistent with other experiments on side-effect actions, which clearly show that subjects do not regard agents as intending to perform or as having the intention to perform those actions. See Pettit and Knobe (2009).

To test this point, I conducted another experiment in which subjects were asked to read the Terrorist Scenario and then answer ‘Yes’ or ‘No’ to the following question: ‘Did the terrorist choose to save the Americans?’ A total of 48 people participated in the survey, the majority of whom (66.7%) answered ‘Yes’.